1 Neuroeconomics of Asset-Price Bubbles
Total Page:16
File Type:pdf, Size:1020Kb
Neuroeconomics of Asset-Price Bubbles: Toward the Prediction and Prevention of Major Bubbles John L. Haracz* Goldman School of Public Policy, UC Berkeley and Department of Psychological and Brain Sciences, Indiana University [email protected] and Daniel J. Acland Goldman School of Public Policy, UC Berkeley [email protected] Goldman School of Public Policy Working Paper January 16, 2015 Abstract Asset-price bubbles challenge the explanatory and predictive power of standard economic theory, suggesting that neuroeconomic measures should be explored as potential tools for improving the predictive power of standard theory. We begin this exploration by reviewing results from functional magnetic resonance imaging (fMRI) studies of lab asset-price bubbles and herding behavior (i.e., following others' decisions). These results are consistent with a neuroeconomics-based hypothesis of asset-price bubbles. In this view, decision making during bubble or non-bubble periods of financial- market activity is driven by, respectively, evolutionarily ancient or new neurocircuitry. Neuroimaging studies that test this or other neuroeconomics-based hypotheses of asset-price bubbles may yield a bubble-related biomarker (e.g., low trade-related lateral neocortical activity associated with traders’ herding-based decisions). Wearable functional near-infrared spectroscopy (fNIRS) technology could determine the prevalence of such a biomarker among financial-market participants, thereby enabling the real-time detection of an emerging bubble. We describe mechanisms by which this early-warning signal could be exploited in self-regulatory or government-administered policies for financial-system stabilization. In summary, neuroimaging-based financial-system regulation may be useful for distinguishing bubbles from non-bubble periods and preventing major asset-price bubbles. *Corresponding author: John L. Haracz, Goldman School of Public Policy, UC Berkeley, 2607 Hearst Ave., Berkeley, CA 94720; phone: (510) 910-2025; email: [email protected] Acknowledgments The authors appreciate comments by seminar participants at Stanford University and Chapman University. Portions of this material were presented at the 2013 and 2014 conferences of the Society for Neuroscience and the Society for Neuroeconomics. 1 I. Introduction It is not as if economic theory has given us the final word on…business cycle and stock market fluctuations. It is hard to believe that a growing familiarity with brain functioning will not lead to better theories for these and other economic domains, perhaps surprisingly soon. Colin F. Camerer, George Loewenstein, and Drazen Prelec (2004, p. 573) Not every puzzle can be solved in the course of normal science; such cases are anomalies. Crisis occurs when a sufficient weight of particularly significant anomalies causes scientists to question the capacity of the current tradition to solve those anomalous puzzles. Alexander Bird (2012, p. 861, italics in original) Econometric methodology encounters difficulties in identifying asset-price bubbles retroactively, let alone in real time (Balke and Wohar, 2009). This difficulty may result from econometricians assigning themselves the hapless task of measuring only external variables when assessing group-level effects (i.e., asset-price bubbles) that arise partly from internal, neuroeconomic processes. The largely missed opportunity to forecast the recent financial-system crisis has led to calls for new macroeconomic theory and methodology (Colander et al., 2009; Stiglitz, 2011, 2014). For example, Janet L. Yellen (2010, p. 243), in a “Closing Panel Presentation” before she became the Chairwoman of the Board of Governors of the Federal Reserve System, questioned the “relevance and usefulness” of the “dynamic stochastic general equilibrium model with nominal rigidities”, which is a model that “ascended to the position of reigning macroeconomic orthodoxy.” From the perspectives of traditional econometrics and macroeconomics, asset-price bubbles could be viewed as unsolved anomalies due to the above analytical and forecasting difficulties. Neuroeconomists now have an opportunity to step into this analytical void left by traditional econometric and macroeconomic approaches. The present review article will assess whether new applications of neuroeconomic methods, in an approach that could be called “neuroeconometrics”, may be useful for detecting the emergence of asset-price bubbles. Early concerns about herding (i.e., decision-making that follows the decisions of others) by William Stanley Jevons, who wrote in The Theory of Political Economy (1871; quoted by De Bondt [2012]) “…as a general rule, it is foolish to do just what other people are doing, because there are almost sure to be too many people doing the same thing”, have been echoed recently due to the potential of herding to destabilize financial markets and yield asset-price bubbles or crashes (Scharfstein and Stein, 1990; Baddeley, 2010; Stiglitz, 2011; Bursztyn et al., 2014). Therefore, neuroimaging signs of herding may be candidate neuroeconometric indicators for an emerging bubble. A high prevalence of these early- warning indicators among market participants could signal regulators to prevent major bubbles and crashes by implementing countercyclical measures (e.g., adjusting caps on loan-to-value ratios for mortgages and increasing capital requirements for banks [Evans, 2011; Bernanke, 2012]). On a voluntary basis, these warning signals also could be used by market participants to trigger self- regulatory actions. To facilitate the development of this neuroimaging-based regulatory approach, we propose a research program that seeks neuroimaging signs of herding associated with decision-making by financial-market participants. Herding-related decisions are hypothesized to drive the emergence of asset-price bubbles, whereas deliberative financial decision-making may be more prevalent during non-bubble periods of financial-market activity. The remainder of the present review article is organized as follows. The following section discusses the distinction between deliberation and herding, while reviewing studies of neural bases for these 2 types of decision making. The section concludes by showing that the neural bases of deliberation and herding provisionally support a novel neuroeconomics-based hypothesis of asset-price bubbles. Section III discusses alternative neuroeconomics-based hypotheses of asset-price bubbles. Sections IV and V, respectively, propose roles for neuroeconomics research in developing self-regulatory or government-administered policies for financial-system stabilization. Section VI introduces neuroeconometrics as a field for studying and applying neuroeconomic measures in the analysis of macroeconomic or financial-market processes. Section VII concludes by emphasizing the welfare effects of elucidating potential neuroeconomic mechanisms underlying asset-price bubbles. II. Deliberation vs. herding …humans may use the same neural machinery to surf the stock exchange that they once used to scavenge the savannah. Brian Knutson and Peter Bossaerts (2007, p. 8176) A. Deliberation Deliberation can include both quantitative, technical analysis and a more qualitative “gist” associated with intuition (Reyna and Huettel, 2014). Experts, even in quantitative fields, rely on intuition (Bird, 2012; Reyna, 2012). For example, studies of financial-market traders show their use of intuition rather than pure technical ability (Fenton-O'Creevy et al., 2005). Lateral neocortical brain regions, particularly lateral prefrontal cortex (PFC) and parietal cortex, show activations in fMRI studies of calculation (Corricelli and Nagel, 2009; Shirer et al., 2012) and gist-related intuition (Reyna and Huettel, 2014). A quantitative meta-analysis of 28 fMRI or positron emission tomography (PET) studies showed predominantly neocortical activations during subjects’ deductive reasoning, which was based on relational, categorical, or propositional arguments (Prado et al., 2011). Brain regions most consistently activated during deductive reasoning included the middle frontal gyrus, inferior frontal gyrus, medial frontal gyrus, precentral gyrus, posterior parietal cortex, and basal ganglia. Intensive reasoning training strengthened fMRI-measured connectivity between lateral PFC and parietal cortex (Mackey et al., 2013). Other deliberative processes associated with activations in lateral PFC and parietal cortex include the exploration of alternative options (Laureiro-Martinez et al., 2014), complex value processing that integrates multiple pieces of information (Dixon and Christoff, 2014), learning to make optimal investment choices (Rudorf et al., 2014), rule-based cognitive control (Dixon and Christoff, 2014), and making comparisons with reference to attributes such as size, numbers, line lengths, angles between lines, luminance, time, beverage taste, physical attractiveness, monetary rewards, and relations between stimuli (Wendelken et al., 2012; Bien and Sack, 2014; Genovesio et al., 2014; Kedia et al., 2014). The degree of frontoparietal activation correlates positively with the similarity of attributes being compared (Bien and Sack, 2014; Kedia et al., 2014) and the computational load imposed by informational uncertainty (Fan et al., 2014). This frontoparietal network, which may have evolved in foraging anthropoid primates to support general problem-solving in humans (Genovesio et al., 2014), can be hypothesized to become activated as asset values